procedure ArCovariance(const Src: TVec; Order: Integer; const a: TVec; out E: Double);
Covariance method for autoregressive parameter estimation.
| # | Name | Type | Description |
|---|---|---|---|
| 1 | Src | TVec | |
| 2 | Order | Integer | |
| 3 | a | TVec | |
| 4 | E | Double |
Result: stored in self (calling object)
Remarks:
The AR parameters are estimated by minimizing an estimate of the prediction error power, but uses less data points then Yull-Walker (autocorrelation method) estimator. The covariance method can accurately extract frequencies of pure sinusoids.
Src contains the data on which the autoregressive parameter estimation (placed in A) should be based. Order defines the Order of the autoregressive process. E is the prediction error.
References:
[1] Modern spectral estimation, Steven M. Kay, Prentice-Hall, Page 221